Difference between pcp_combine -sum and -add

[BinbinZhou-NOAA]

What is the difference between pcp_combine's -sum and -add options
For only 2 input files, they are same?

[JohnHalleyGotway]

@BinbinZhou-NOAA, my answer to your question is kind of, but let me explain. While its true that running -add for two input files will produce the exact same numeric results as running -sum to locate and add them, the command line arguments are very different.

The -sum option uses:

	SUM_ARGS:
		init_time
		in_accum
		valid_time
		out_accum

While the -add option uses:

	INPUT_FILES:
		file_1 config_str_1 ... file_n config_str_n

For some background, the -sum option was written first many years ago. It only works with input data with a consistent accumulation interval, such as 3-hour accumulations of precipitation. It uses the command line arguments to decide what data to look for... dividing the requested out_accum by the in_accum to figure out the number of files to find. I searches through the input -pcpdir directory(ies), processing each file it finds that matches the -pcprx regular expression, looking for precipitation (APCP) data that matches the one of the timesteps.

Whereas the newer -add option is much simpler. It just reads the requested data from the files specified on the command line.

The following 2 commands produce the same result:

pcp_combine -sum 20050807_000000 3 20050807_060000 06 sum_06a.nc -pcpdir data/sample_fcst/2005080700

and

pcp_combine -add data/sample_fcst/2005080700/wrfprs_ruc13_03.tm00_G212 data/sample_fcst/2005080700/wrfprs_ruc13_06.tm00_G212 -field 'name="APCP"; level="A03";' add_06a.nc

The -sum option is limited to only processing "accumulating" variables, like APCP. And I think it only works well with GRIB inputs. It's also slow because it has to go open up many files looking for the matching data.

The -add option is faster since it only opens the exact files that it needs. The harder part is figuring out which files to pass to it... but that's the advantage of METplus. The METplus wrapper will build the call to pcp_combine for you, as long as you've configured it well.

[BinbinZhou-NOAA]

I also observed the the -sum option is very slow. Based on your explanation, -add can replace -sum option.
and will run faster.

Thanks!

[JohnHalleyGotway]

@BinbinZhou-NOAA, great, glad that explanation help. Yes, I would also advise using -add instead of -sum for the reasons you've already found. I'll go ahead and mark this discussion as being answered and also "lock it" to prevent future posts.

As always, feel free to create new discussions when new issues or questions arise.